Oxford Review of Economic Policy, Volume 37, Number 4, 2021, pp.

824–837

Fixing capitalism’s good jobs problem

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

Dani Rodrik* and Stefanie Stantcheva**

Abstract Conventional welfare state policies that centre on education, training, progressive taxation,
and social insurance are inadequate to address labour market polarization, which is capitalism’s most
pressing inclusion challenge at present. We propose a strategy aimed directly at the productive sphere
of the economy and targeting an increase in the supply of ‘good jobs’. The main elements of this
strategy are: (i) active labour market policies linked to employers; (ii) industrial and regional policies
directly targeting the creation of good jobs; (iii) innovation policies that incentivize labour-friendly
technologies; (iv) international economic policies that facilitate the maintenance of high domestic la-
bour/social standards. These elements are connected both by their objective—expanding the number
of good jobs—and by a new approach to regulation that is collaborative and iterative rather than top-
down and prescriptive. We emphasize the importance of new institutional arrangements that enable
strategic long-term information exchange and cooperation between governments and firms.
Keywords: good jobs, welfare state, reforming capitalism
JEL classification: D60, F13, H10, H20

I. Introduction
Capitalism is not a rigid system. It has evolved and changed over time, shaped by local
history, social pressures, and crises. Today, it is commonplace to contrast the ‘state
capitalism’ of China with the less state-directed version in the West. But even within
Europe and North America, there are significant differences among Nordic corporatist
and the Anglo-American variants of the market economy. And all these contemporary
models of capitalism differ from the historical versions that prevailed until the emer-
gence of the welfare state after the 1930s.
Markets are not self-creating, self-regulating, self-stabilizing, or self-legitimizing.
Hence, every well-functioning market economy relies on non-market institutions to
fulfil these roles. Often put in place by the state, these institutions range from prop-
erty and contract enforcement systems to social insurance and income redistribution
mechanisms. Capitalism’s adaptability derives from the inherent plasticity of such

*Harvard University, USA, e-mail: dani_rodrik@hks.harvard.edu

**Harvard University, USA, e-mail: sstantcheva@fas.harvard.edu
This paper draws heavily on Rodrik and Sabel (2021) and Rodrik and Stantcheva (2021). We thank Derek
Morris, the editors, and participants at the Oxford Review’s editorial seminar for useful comments.
https://doi.org/10.1093/oxrep/grab024
© The Author(s) 2021. Published by Oxford University Press.
For permissions please e-mail: journals.permissions@oup.com
Fixing capitalism’s good jobs problem 825

arrangements. Even property rights, typically considered the cornerstone of a market

economy, can be repackaged and reconfigured in an almost infinite variety of ways.1
What counts as property and where its boundaries lie are questions to which there have
not been immutable answers.
The malleability of capitalism is its great strength. Crises of capitalism are as old as

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

capitalism itself. Yet, each time, capitalism has survived, reforming and adapting itself
to new challenges.
The two biggest challenges today’s capitalism faces are climate change and social
inclusion. In its own way, each of these two challenges is existential. Climate change
threatens humanity through its potentially catastrophic consequences for our physical
environment. The economic, cultural, and spatial divides that have deepened within
countries in recent decades, on the other hand, threaten the viability of our societies
and polities.
In this paper, we address the second of these challenges. We argue that the con-
ventional welfare state policies centring on education, training, progressive taxation,
and social insurance are on their own inadequate to address today’s inclusion chal-
lenge. We propose a multi-pronged strategy aimed directly at the productive sphere
of the economy and targeting an increase in the supply of ‘good jobs’—jobs that
provide a middle-class living standard, a sufficiently high wage, good benefits, reason-
able levels of personal autonomy, adequate economic security, and career ladders.2
The main elements of this strategy are: (i) active labour market policies linked to
employers; (ii) industrial and regional policies directly targeting the creation of good
jobs; (iii) innovation policies that incentivize labour-friendly technologies; (iv) inter-
national economic policies that facilitate the maintenance of high domestic labour/
social standards.
These elements are connected both by their objective—expanding the number of
good jobs—and by a new approach to regulation that is collaborative and iterative
rather than top-down and prescriptive. We emphasize the importance of new insti-
tutional arrangements that enable strategic long-term information exchange and co-
operation between governments and firms (as well as between the general public and
policy-makers)—underscoring the need for joint endeavour by public, private, and
civic sectors to deliver improvements for all. In the rest of the paper, we discuss this
approach and illustrate it in the context of the four specific elements of the good-jobs
strategy.3

1 It is useful in this context to consider how China has managed to provide effective property rights in-
centives and contract enforcement through institutional arrangements that differ greatly from conventional
Western ones (Rodrik, 1996; Qian, 2017).
2 Osterman (2020) provides a very valuable set of case studies on what can be done to increase good jobs

in specific industries. See also Ton (2014), who discusses how it may be in many firms’ self-interest to enhance
the quality of the jobs they offer.
3 This paper focuses on advanced economies. Many developing economies seemed to have been doing

better prior to the pandemic, on the back of export-oriented industrializing strategies. However, even before
the pandemic there were signs that their growth models were running out of steam (see Diao et al., 2019, and
Rodrik, 2018).
826 Dani Rodrik and Stefanie Stantcheva

II. Stating the problem and the general approach

One of the fundamental problems of contemporary capitalism is its failure to produce
adequate numbers of good jobs to sustain a prosperous and growing middle class.
Quantitative indicators of this can be found in measures of labour market polariza-

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

tion, rising spatial inequality, declining job stability, greater self-reported economic in-
security, and declining middle-class income shares (Autor and Dorn, 2013; Eurofound,
2017; OECD, 2019). Figure 1 shows the collapse in the number of jobs for medium-skill
occupations in leading advanced economies. This figure is taken from OECD (2019)
and similar results can be found in studies by Autor et al. (2020) and McKinsey Global
Institute (2020). Even in countries (such as France) where there does not seem to have
been a marked deterioration in overall inequality, these labour market syndromes are
acutely felt (Rodrik and Stantcheva, 2021).
The disappearance of good jobs is not just an equity problem. It is a broader social
problem since declining labour market opportunities produce a wide variety of social
ills such as family breakdown, crime, and substance abuse (Case and Deaton, 2020). It
is also a political problem as it fosters the rise of authoritarian and nativist populism
(Rodrik, 2021). It is even a problem for economic efficiency and growth insofar as it
slows down the dissemination of innovation from the more advanced sectors and firms
to the rest of the economy through the creation of more productive jobs in the middle
of the skill distribution.
How should we deal with this problem? It is useful to consider our options with the
help of a matrix that categorizes different types of remedies for inequality (Figure 2).
First, we divide policies into pre-production, production, and post-production stage
interventions. For our purposes, this is a better categorization of policies than the more
conventional pre-distribution versus redistribution distinction. It distinguishes between
policies that affect endowments people bring to markets (such as education and skills)
and policies that influence production, employment, and investment decisions (such as

Figure 1: Labour market polarization

Source: OECD (2019).

Fixing capitalism’s good jobs problem 827

Figure 2: Remedies for inequality

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

Notes: UBI is universal basic income. EITC is earned income tax credit.

industrial policies or labour-market regulations). Second, we divide interventions into

those that intend to redress inequities at the bottom, middle, or top of the income dis-
tribution. Minimum wages, for example, target the incomes of the working poor, while
wealth taxes target those at the very top. It is possible to fill all nine cells of the table
with examples of contemporary policies, as is done in Figure 2.
The matrix clarifies the differences between alternative approaches to equity and in-
clusion. The current policy discussion in the US, for example, focuses almost exclusively
on the first and third rows of the matrix: how to address poverty at the bottom and
concentration of income at the very top. The traditional welfare state model operates
largely within the first and third columns, too: it targets the educational and other en-
dowments of workers before they join labour markets on the one hand and ex post re-
distribution through taxes and social insurance policies on the other. The government’s
role is to finance education, engage in progressive taxation, and provide social insur-
ance against idiosyncratic risks such as unemployment, illness, and disability. In either
case, the assumption is that good middle-class jobs will be available to all with adequate
education and skills.
We need a different—or at least complementary—approach when inequality and
economic insecurity are structural problems, and when the inadequacy of good middle-
class jobs is driven by secular trends such as technology and globalization. Preparing
young workers for the labour market and reskilling older workers for newer occupa-
tions will not work if firms are not supplying an adequate quantity of good jobs. This
calls for targeting the middle cell of the matrix, focusing on direct interventions in the
productive sphere with the goal of expanding the supply of middle-skill jobs.
We discuss what such a good-jobs strategy might look like under four separate head-
ings below. Two points are worth making at the outset. First, good jobs cannot be
828 Dani Rodrik and Stefanie Stantcheva

generated by fiat; they are contingent on higher productivity and the expansion of good
firms. The proposed strategy is one that specifically focuses on productivity enhance-
ments along the middle spectrum of the labour market. Second, the strategy requires
an approach to governance and regulation that differs from the standard ex ante, arm’s
length model that is familiar to economists. We elaborate on that governance model

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

after discussing the specific elements of the strategy.

III. Active labour market policies linked to employers

Active labour market policies are programmes that aim to increase the beneficiaries’
prospect of finding employment or increasing earnings. European programmes in this
domain include skills training and certification, employment subsidies, public-sector
employment, and assistance with job search. Studies on their impacts have generally
yielded mixed results (Caliendo and Schmidl, 2016). However, a particular approach to
skills training, called ‘sectoral training programs’ in the US, has produced much more
encouraging results.
Sectoral programmes differ from general training courses in that they are oriented to-
wards the need of particular employers and require much greater cooperation with them.
They also provide a wider range of customized services to job-seekers. Exemplified by
Project Quest in San Antonio, Texas, they entail: training in soft skills as well as occupa-
tion-specific skills and credentials; partnerships with community colleges and employers;
extensive wraparound and follow-up services in addition to training and job placement;
and a dual-customer approach that involves employers as well as job-seekers. Sectoral
training programmes have been evaluated repeatedly through randomized methods and
have been shown to produce significant and sustained gains for participants (Maguire
et al., 2010; Schaberg, 2017; Roder and Elliott, 2019; Katz et al., 2020).
Key to their success is their targeting of specific industries or occupations that have
the potential to create more local employment. Programme staff work closely with
employers, and the firms themselves may serve on the programmes’ boards. Training
courses are designed in close cooperation with prospective employers. Strong links with
labour unions and local governments help, too, as these provide additional vehicles
through which workers can be placed. As the relationship develops, employers start to
see these programmes as an important asset. Since firms benefit from the training, they
are willing to cooperate with the programme and sometimes even adjust their employ-
ment practices. Increased trust between employers and training agencies can pay off in
the form of higher productivity for the firm as well as increased labour market oppor-
tunities for job-seekers.
Despite their success, such programmes have remained very small and limited in
scope.4 In the US, they are associated with non-governmental civic or community or-
ganizations rather than state agencies. In Europe, there are extensive networks of Public
Employment Services (PES) which, however, typically operate at arm’s length from em-
ployers and provide limited services to workers. More recently, some PES (in Germany
and Sweden, for example) have experimented with greater employer-orientation.

4 President Biden’s infrastructure plan commits federal resources to scaling up sectoral training pro-

grammes, and is an important first step.

Fixing capitalism’s good jobs problem 829

There is an opportunity to enhance training and active labour market policies by

transforming them into vehicles of sustained engagement and collaboration with local
(and prospective) employers. The provision of customized services to both job-seekers
and employers can help achieve multiple objectives. It makes training and job place-
ment more effective. It makes it possible to reach workers who might otherwise drop

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

out of the labour market because of particular circumstances (such as lack of trans-
port, child care, or specific gaps in education). It might not only enhance the product-
ivity of local firms through qualified employees, but also induce them to adapt their
employment and human resources practices to the needs of local labour markets.
Ultimately, the full quid pro quo—more good jobs in return for more good workers—
can be realized only when businesses recognize the benefits of the services provided
to them by public-sector training and placement agencies. Therefore, moving beyond
placement to productivity requires not just the right institutional designs but also a pro-
cess of building trust among social partners—employers, workers’ organizations, and
public agencies such as the PES. Developing the requisite social capital will necessarily
take time. It will also require complementary interventions in the areas of industrial
and innovation policies, to which we turn next.

IV. Industrial and regional policies targeting the creation

of good jobs
Despite economists’ apparent aversion to ‘industrial policies’, the latter have always
been part of most governments’ arsenals, simply changing shape and focus (and, some-
times, just names) as economic priorities and fashions evolved. In the US, the practice
of industrial policy has a long history, even if the term has carried a note of disrepute
until very recently. It has taken a wide range of forms—from the Defense Advanced
Research Projects Agency (DARPA) to Small Business Administration programmes, to
widespread state-level business incentives. In Europe, even though state aid is generally
frowned upon, investment incentives for small businesses and lagging regions remain
rampant. In recent years, the need for industrial policy has been articulated more ex-
plicitly and forcefully. The challenges of transition to a green economy, geographic
divides, digitalization, and the perceived threat of Chinese competition in high-tech
industries have highlighted the urgency of public action to stimulate investment and
innovation in particular industries and regions.
While employment creation is almost always a subsidiary goal of these programmes,
industrial incentives are rarely designed around the imperative of good jobs. For ex-
ample, in what is one of the best studies of industrial policy, a paper by Criscuolo et al.
(2019) analyses the effects of the Regional Selective Assistance (RSA) programme in
Britain. The programme aimed to maintain and expand employment in low-income,
high-unemployment areas, but was essentially a subsidy to capital. Firms could apply
to the government with specific investment plans, either to finance new capital equip-
ment or to modernize existing plants. If approved, the government financed up to 35
per cent of the investment. The programme did have a significant effect on employment
in smaller firms according to Criscuolo et al. (2019), but it did not directly subsidize
good jobs or other activities that may have had a more direct impact on jobs.
830 Dani Rodrik and Stefanie Stantcheva

A second consideration is that business incentives work best when they are custom-
ized and targeted to specific needs of firms, and when they are part of an iterative dia-
logue between firms and government agencies. The traditional caricature of industrial
policy, in which bureaucrats choose a set of economic activities to be promoted, select
pre-determined incentives (e.g. tax rebates or subsidized credit), and then impose hard

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

conditionality on the receiving firms does not accurately characterize how industrial
policy was actually implemented in Japan, Taiwan, South Korea, or China. Successful
programmes tend to revolve around a process of strategic collaboration, in which firms’
needs, market opportunities, and appropriate remedies are discovered over time, with
policies revised as learning takes place.
Tim Bartik of the Upjohn Institute has been a long-term observer of business incen-
tives in the US, and his synthesis of the evidence provides a valuable perspective that can
apply equally well to other contexts (Bartik, 2019, 2020). The current approach revolves
largely around significant tax breaks that often go to large corporations and are not
properly targeted or designed. They also involve a very large cost per job created. Bartik
makes several recommendations. First, business incentives should focus on areas that
are distressed—i.e. areas that truly need them. Second, the incentives should focus on
sectors or firms that are likely to have high job creation potential. Third, public assist-
ance should focus less on tax incentives (and on encouraging physical investment) and
more on specific public services needed by firms, such as customized business services,
zoning or infrastructure policies, local amenities, and skills training. Fourth, business
assistance should be viewed as a portfolio of services rather than a particular incentive,
with the actual mix attuned to local conditions.
Bartik’s recommendations echo ideas that have developed over the last couple of
decades into a new conception of industrial policy (Evans, 1995; Rodrik, 2007, 2008;
Sabel, 2007; Hausmann et al., 2008; Fernández-Arias et al., 2016; Ghezzi, 2017). Under
this conception, the government is not presumed to know where the market failures are
beforehand and, therefore, does not determine ex ante what the specific policy instru-
ments are. Industrial strategy consists of a collaborative process of ‘discovery’ involving
business and agencies of the state, where the objective is to identify the constraints and
opportunities over time, and to design interventions appropriately. As learning takes
place, policies are revised, refined, and sometimes reversed. We know such practices are
feasible because they already exist in a number of policy domains; Rodrik and Sabel
(2021) discuss water-quality regulation in Europe and promotion of high-tech innov-
ation through DARPA in the US, while Ghezzi (2017) discusses their application to
modern agriculture in Peru.
In return for services provided by state agencies, firms would be asked to make provi-
sional commitments on specific quantities of jobs they will create at different qualifica-
tion levels (i.e. low salaried employees, medium-salaried employees, etc.). Firms would
be encouraged to pool proposals when they make use of common inputs—as would
be the case for workers with particular skills or infrastructure. Other conditionalities
might be included as well. A firm might be asked to work with local suppliers to im-
prove their management or technological capabilities. Or a firm that is considering out-
sourcing part of its production to a foreign country may be asked to delay doing so for
a number of years, in case productivity improvements at home may render those plans
unnecessary. The firm may be required to arrange for additional training for some of
its employees.
Fixing capitalism’s good jobs problem 831

A particular objective would be to connect business incentives more tightly with the
kind of labour market and training programmes discussed previously. While labour
market interventions may do a good job of preparing job-seekers for good jobs, their
effects will remain limited if there is not a corresponding increase in the supply of good
firms and of good jobs in existing firms.

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

V. Innovation policies that incentivize labour-friendly
technologies
In 2016 Elon Musk announced that Tesla’s Model 3 would be built in a new, fully auto-
mated car factory. Codenamed ‘Alien Dreadnought’, the project would allow robots in
the factory to operate beyond human speed, with raw materials coming in at one end
and finished cars rolling out the other. By mid-2018 it was clear that automation was
not working as expected and Tesla was driven close to financial collapse. The company
was forced to launch a new assembly line full of human workers. Musk would say later:
‘people are way better at dealing with unexpected circumstances than robots.’5 ‘Yes, ex-
cessive automation at Tesla was a mistake. . . . Humans are under-rated,’ he conceded
on Twitter.
Tesla’s automation mistake is revealing for several reasons. First, it highlights how
production techniques relying on human labour can still dominate automation when it
is impossible to fully account for uncertainty and routinize all tasks. Second, it is indica-
tive of the excessive faith many business leaders often place on new technologies. Third,
it reminds us that technology adoption is a choice: businesses face a range of options
about what kind of innovations to use and deploy—choices that have significant im-
plications for the workforce—and sometimes, society as a whole—but are not typically
internalized in the decision-making process.
The usual discussion around the labour market implications of new technologies is curi-
ously one-sided. The direction of technological change—whether it augments or replaces
labour—is taken to be essentially pre-determined and out of our control. It is workers and
society at large that have to adjust to technological change—rather than the other way
around. But as the late Anthony Atkinson emphasized, the determination of the direction of
technological change cannot be left to firms and innovators alone (Atkinson, 2015, pp. 115–
18). This argument has been picked up more recently by Daron Acemoglu (Acemoglu and
Restrepo, 2019) (see also Korinek (2019) and Rodrik and Stantcheva (2021)).
The direction of technological change depends on several conditions that may be
amenable to policy influence. First and most directly, government-funded and -directed
innovation programmes make decisions about what kind of innovations to promote.
Defence-related and green technologies are examples. Employment-friendly technolo-
gies—those that augment rather than replace labour—could be part of those priorities,
though they are not at present. Second, private-sector innovation incentives can be
skewed because of prevailing financing methods or policies. Venture capital, for ex-
ample, naturally seeks areas where the returns can be capitalized relatively quickly by

5 https://www.teslarati.com/inside-tesla-tent-based-model-3-production-assembly-line-profitability/
832 Dani Rodrik and Stefanie Stantcheva

investors. This may exclude innovations where the gains are longer term or reaped by
society at large (Lerner and Nanda, 2020).
Third, most advanced economies tax capital income more lightly (through depreci-
ation allowances and various incentives of the type we discussed previously) and tax
labour more heavily (through personal income taxes, social insurance contributions,

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

and payroll taxes). An unintended consequence of such a tax system is to make it more
attractive to firms to economize on labour by investing in machinery, to an extent that
may be socially suboptimal (Acemoglu et al., 2020).
Finally, the direction of technological change also depends on the balance of power
between employers and employees. When workers have a say in the workplace, manage-
ment has to get buy-in from them before major technologies are deployed and work is
restructured. This can result in a modern version of Luddism—aversion to any kind of
innovation that appears to threaten jobs. But it can also be a useful counterweight to
adverse incentives in the system encouraging too much automation or the adoption of
what Acemoglu et al. (2020) call ‘so-so’ technologies.6 In short, the direction of tech-
nical change, in addition to its rate, depends on a wide range of factors, many of which
could be influenced by societal and governmental decision-making.
As a matter of logic, the gap between skills and technology can be closed in one of
two ways: either by increasing education to match the demands of new technologies,
or by redirecting innovation to match the skills of the current (and prospective) labour
force. The second strategy, which gets practically no attention in policy discussions, is
worth taking seriously. It may be possible to direct technology to better serve the ex-
isting workforce’s needs, in addition to preparing the workforce to match the require-
ments of technology.
Acemoglu and Restrepo (2018) have argued that it is possible to countervail present
technological trends and push innovation in a direction that creates new, labour-aug-
menting tasks. They cite three areas. First, they suggest artificial intelligence (AI) could
be used in education in order to create more specialized tasks for teachers, personalize
instruction for students, and increase effectiveness of schooling in the process. Second,
they note a similar potential in healthcare, which is perhaps closer to realization. AI
tools can significantly enhance the diagnostic and treatment capabilities of nurses, phys-
icians’ aides, and other medical technicians, allowing a priori ‘less skilled’ practitioners
to perform tasks that only physicians with many more years of professional educa-
tion have traditionally undertaken. Third, Acemoglu and Restrepo (2018) mention the
use of augmented and virtual reality technologies in manufacturing, enabling humans
and robots to work together in performing precision tasks (rather than the latter re-
placing the former). Such technologies are based on smaller, more nimble robots that
also enable greater customization of production in response to specific customer needs.
Indeed, companies such as BMW and Mercedes are building their automation plans
around human work, which they have found allows both for greater reliability and more
customization in production.
These considerations suggest some broad directions for policy. Prevailing fiscal re-
gimes can be reviewed to ascertain whether there are excessive incentives for investment

6 ‘So-so’ technologies are those that substitute labour, but are not that much more productive at the task

than workers are, thus not increasing productivity sufficiently.

Fixing capitalism’s good jobs problem 833

in automation. Employment considerations can be incorporated directly in the existing

regime of tax incentives for R&D. The selection criteria could revolve around the mar-
gins of choice we discussed previously. ‘So-so’ innovations that directly replace labour
without significant overall productivity benefits would be favoured the least, and innov-
ations that augment labour of low and medium skills and create new, labour-absorbing

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

tasks would be favoured the most. Governments could apply a ‘prospective employ-
ment test’ when determining their public-spending priorities for innovation. In the EU,
for example, employment considerations appear to play virtually no direct role in the
construction of the innovation portfolio or in the design of the European Green Deal.
One possibility, among many, is to devote a portion of the European Fund for Strategic
Investments (EFSI) to experiment in developing labour-friendly technologies.
In addition, governments can directly encourage the introduction and dissemination
in the private sector of learning organizations that empower workers. The goal would
be for such organizational forms—based on teamwork, development of cognitive, so-
cial, and soft skills, workers’ autonomy, and continuous learning—to replace Taylorist
or lean organizational models where feasible. Finally, public policy can play a role in
shaping public consciousness about the social and employment consequences of innov-
ation. A public that is more aware about the choices we have is likely to expect more
from innovators. It would be highly desirable to have a shift in the public narrative
on technology that recognizes innovation can be directed in a more labour-friendly
direction.

VI. International economic policies that safeguard high

domestic labour/social standards
Are the policies of the sort we have discussed here that induce domestic firms to expand
good jobs feasible in a globalized economy? What if firms evade domestic responsibil-
ities by outsourcing, hiding in tax shelters, or if they lose competitiveness and market
share to firms in countries with lower standards? The principal safeguard against such
a race to the bottom is that the good-jobs strategy is as much about enhancing prod-
uctivity—especially for lagging firms—as it is about jobs. It recognizes that good jobs
require good firms. Enhanced productivity makes domestic firms better at global com-
petition and should reduce incentives for outsourcing and moving abroad. Therefore,
there is no inherent conflict between the good-job strategy and the open economy.
Nevertheless, such policies can be buttressed by supportive international economic
policies. One approach is to negotiate international agreements that uphold higher
standards. Greater international cooperation and information exchange on taxation
of corporations and wealthy individuals would help. So would the incorporation of
enforceable labour-rights provisions in international trade and investment agreements.
But in view of the heterogeneity among countries and the limitations of enforcement
on sovereign nations, international standard-setting on taxes and labour regulations
can go only so far.
A complementary approach would be to incorporate in domestic trade policy an
explicit safeguard mechanism for addressing imports that threaten to undermine do-
mestic social and labour standards. One of us has described elsewhere an anti-social
834 Dani Rodrik and Stefanie Stantcheva

dumping procedure designed to achieve that objective, which we summarize here

(Rodrik, 2019). Explicit ‘safety valves’ allowing countries to raise trade barriers under
certain conditions is a means for enhancing the legitimacy of international trade and
outsourcing in general. This is a principle already embodied in ‘fair trade’ provisions
of trade agreements.

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

When international trade operates just like any domestic form of market competi-
tion, it makes little sense to set it apart and treat it differently from other approaches
for dealing with inequality and insecurity in labour markets at large (using unemploy-
ment compensation, progressive tax systems, active labour market policies, employ-
ment-friendly macro policies, etc.). But when trade entails practices that violate laws
or norms embodied in domestic institutional arrangements, and thereby undercuts do-
mestic social bargains, it may be more legitimate to restrict the import flows that have
the alleged effect. Restrictions on imports should not be permissible merely because
wages in an exporting country are low. But trade may be considered unfair when com-
petitive advantage is gained through the violation of worker rights.
A policy that targets social dumping must distinguish between true social dumping
and ordinary market competition. Therefore, it needs a domestic investigatory pro-
cess of fact finding, as in the case with regular anti-dumping. The investigative pro-
cess in each country would: (i) determine that the imports in question do threaten to
undermine a domestic standard or widely held social norm; (ii) gather public testimony
and views from all relevant parties, including consumer and public-interest groups, im-
porters of the product(s) concerned, and exporters to the affected country; and (iii)
ascertain whether there exists broad support among these groups for the application of
the safeguard measure in question.
Ordinary protectionism would not have much chance of success if groups whose
incomes would be adversely affected by trade restrictions—importers and exporters—
were necessarily part of the deliberative process and the investigative body had to deter-
mine whether these groups also support the safeguard measure. At the same time, when
deeply and widely held social norms are at stake, these groups are unlikely to oppose
safeguards in a public manner, as this would endanger their standing among the public
at large. Imagine, for example, that forced labour was used in producing goods for ex-
port in country X, or that labour rights were widely and violently repressed. Exporters
to country X and downstream users of X’s products may find it difficult to publicly
defend free trade with this country.
In less clear-cut cases, the main advantage of the proposed procedure is that it would
force a public debate on the legitimacy of trade and when it may be appropriate to
restrict it. It would be incumbent on governments to ensure that the requirements of
democratic deliberation are fulfilled. Are the views of all relevant parties, including
consumer and public-interest groups, importers and exporters, civil society organiza-
tions, sufficiently represented? Is all relevant evidence, scientific and economic, brought
to bear on the final determination? Is there broad enough domestic support in favour of
the opt-out or safeguard in question? These procedural requirements echo those in the
existing World Trade Organization Agreement on Safeguards, although the scope of its
application would be greatly enlarged.
This procedure would force a deeper and more representative public debate on the
legitimacy of trade rules and on the conditions under which it may be appropriate to
Fixing capitalism’s good jobs problem 835

suspend them. The most reliable guarantee against abuse of opt-outs is informed de-
liberation by the polity at large. The requirements that groups whose incomes would be
adversely affected by the opt-out—importers and exporters—participate in the delib-
erations and that the domestic process balance the competing interests in a transparent
manner would minimize the risk of protectionist measures benefiting a small segment

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

of industry at large cost to society. A safety valve that allows principled objections to
free trade to prevail makes it easier to repress protectionist steam. A deepening back-
lash against trade may, in fact, be rendered more likely in the absence of a clause against
social dumping.

VII. A new model of governance

The regulatory model that underlies the good-jobs strategy we have outlined here dif-
fers from the standard, arm’s length regulation model of economists. In the conven-
tional approach, regulatory agencies set explicit guidelines that regulated entities have
to meet, and consultation between the regulator and those entities is limited typically
to resolving differences. There are fixed limits on permissible behaviour and a schedule
of fines for violating them. This model does not apply well to circumstances with high-
dimensional uncertainty.
In the present context, the objective itself (‘good jobs’) is imprecise and multi-
dimensional; it needs to be operationalized in a way that is both evolving and
context-dependent. Furthermore, creating good jobs depends on a wide array of deci-
sions on investment, technological choice, and business organization, the consequences
of which are unknowable ex ante. Technological and operational possibilities are highly
uncertain, and neither firms nor government agencies have the information needed to
devise concrete behavioural schedules from the outset. Hence the interaction between
the government and firms must take as its starting point the provisionality of ends and
means and the need for disciplined review and revision. Targets and instruments for
good-job creation must remain provisional, to be revised as new information comes
in. The task of governance is to establish an information exchange regime that induces
firms and other entities to cooperate with the government and adjust their strategies in
the desired direction in a context of extreme uncertainty.
Such iterative, collaborative practices have become established in industries as diverse
as biotechnology, IT, and advanced manufacturing, as well as in policy regimes such as
food safety, water quality, civil aviation, and the promotion of advanced technologies
(Rodrik and Sabel, 2021). And similar practices abound in successful programmes in la-
bour markets, business incentives, and technology promotion, even though these do not
typically constitute explicit and self-conscious departures from the standard regulatory
model. Many state agencies have considerable experience of working closely with small
and medium-sized enterprises (SMEs), using a wide range of instruments (loans, guar-
antees, equity participation, export credits, training, management counselling, access
to technology and networks). Development banks have the capacity to screen firms,
monitor their progress, and intervene at various stages of their lifecycle. Successful re-
gional investment incentive programmes, such as the RSA mentioned previously, entail
considerable discretion on the part of state agencies, and require significant monitoring
and ongoing negotiations with private firms.
836 Dani Rodrik and Stefanie Stantcheva

VIII. Concluding remarks

We conclude by mentioning four specific advantages of the proposed approach. First,
it is explicitly and self-consciously structuralist. It aims to shape production, innov-
ation, employment incentives, and relationships in situ, instead of taking them as given

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

and only addressing their consequences after the fact, through taxes and transfers. As
such, it advocates a transition from the traditional welfare state to a productivist and
innovation-driven welfare state. Second, it moves us away from the institutional fet-
ishism reflected in the traditional distinction between markets and states. It pushes us to
think in terms of complementarities between private actors and state agencies and col-
laborative, iterative rule-making under extreme, multi-dimensional uncertainty. Third,
it does away with the equity versus growth dichotomy or trade-off: economic growth
is possible only through the diffusion of good, productive jobs from the advanced to
the lagging sectors and regions of the economy; good jobs are possible only if there are
productive firms. Finally, the general approach to regulatory governance highlighted in
the paper is open ended and can be applied within different domains and at different
levels of the economy and society. Potentially it opens up a path of radical institutional
reform from gradualist beginnings. Today’s reforms may eventually cumulate to a revo-
lutionary transformation of the capitalist system.

References
Acemoglu, D., and Restrepo, P. (2018), ‘Artificial Intelligence, Automation and Work’, NBER Working
Paper Series, no. 24196, January, https://doi.org/10.3386/w24196.
— — (2019), ‘The Wrong Kind of AI? Artificial Intelligence and the Future of Labor Demand’,
Cambridge Journal of Regions, Economy and Society, 13(1), 25–35, https://doi.org/10.1093/cjres/
rsz022.
—     Manera, A., and Restrepo, P. (2020), ‘Does the US Tax Code Favor Automation?’, Brookings
Papers on Economic Activity, BPEA Conference Drafts, 19 March.
Autor, D. H., and Dorn, D. (2013), ‘The Growth of Low-skill Service Jobs and the Polarization of the
US Labor Market,’ American Economic Review, 103(5).
—     Mindell, D., and Reynolds, E. (2020), The Work of the Future: Building Better Jobs in an Age of
Intelligent Machines, MIT Work of the Future Report, November.
Atkinson, A. B. (2015), Inequality: What Can Be Done?, Cambridge, MA, Harvard University Press.
Bartik, T. J. (2019), ‘Should Place-based Jobs Policies Be Used to Help Distressed Communities?’
Kalamazoo, MI, W.E. Upjohn Institute for Employment Research, 1 August.
— (2020), ‘Bringing Jobs to People: Improving Local Economic Development Policies’, Aspen
Institute, August.
Caliendo, M., and Schmidl, R. (2016), ‘Youth Unemployment and Active Labor Market Policies in
Europe’, IZA Journal of Labor Policy, 5(1).
Case, A., and Deaton, A. (2020), Deaths of Despair and the Future of Capitalism, Princeton, NJ,
Princeton University Press.
Criscuolo, C., Martin, R., Overman, H. G., and Van Reenen, J. (2019), ‘Some Causal Effects of an
Industrial Policy’, American Economic Review, 109(1), 48–85.
Diao, X., McMillan, M., and Rodrik, D. (2019), ‘The Recent Growth Boom in Developing Economies:
A Structural-change Perspective,’ in M. Nissanke and J. A. Ocampo (eds), The Palgrave Handbook
of Development Economics: Critical Reflections on Globalization and Development, Palgrave
Macmillan.
Fixing capitalism’s good jobs problem 837

Eurofound (2017), Income Inequalities and Employment Patterns in Europe Before and After the Great
Recession, Luxembourg, Publications Office of the European Union.
Evans, P. B. (1995), Embedded Autonomy: States and Industrial Transformation, Princeton, NJ,
Princeton University Press.
Fernández-Arias, E., Sabel, C., Stein, E. H., and Trejos, A. (2016), Two to Tango: Public–Private
Collaboration for Productive Development Policies, Washington, DC, Inter-American Development

Downloaded from https://academic.oup.com/oxrep/article/37/4/824/6423487 by MIT Libraries user on 19 January 2022

Bank.
Ghezzi, P. (2017), ‘Mesas Ejecutivas in Peru: Lessons for Productive Development Policies’, Global
Policy, 8(3), 369–80.
Hausmann, R., Rodrik, D., and Sabel, C. (2008), ‘Reconfiguring Industrial Policy: A Framework
with an Application to South Africa’, August, https://papers.ssrn.com/sol3/papers.
cfm?abstract_id=1245702
Katz, L. F., Roth, J., Hendra, R., and Schaberg, K. (2020), ‘Why Do Sectoral Employment Programs
Work? Lessons from WorkAdvance’, NBER Working Paper, December.
Korinek, A. (2019), ‘Labor in the Age of Automation and Artificial Intelligence’, EfIP Policy Brief,
February.
Lerner, J., and Nanda, R. (2020), ‘Venture Capital’s Role in Financing Innovation: What We Know and
How Much We Still Need to Learn’, Journal of Economic Perspectives, 34(3), 237–61.
Maguire, S., Freely, J., Clymer, C., Conway, M., and Schwartz, D. (2010), Tuning in to Local Labor
Markets: Findings from the Sectoral Employment Impact Study, Philadelphia, PA, Public/Private
Ventures.
McKinsey Global Institute (2020), The Social Contract in the 21st Century, 5 February.
OECD (2017), OECD Employment Outlook 2017, Paris, OECD Publishing, https://doi.org/10.1787/
empl_outlook-2017-en.
—     (2019), The Future of Work: OECD Employment Outlook 2019, Paris, OECD Publishing.
Osterman, P. (2020), Creating Good Jobs: An Industry-based Strategy, Cambridge, MA, MIT Press.
Qian, Y. (2017), How Reform Worked in China: The Transition from Plan to Market, Cambridge, MA,
MIT Press.
Roder, A., and Elliott, M. (2019), ‘Nine Year Gains: Project QUEST’s Continuing Impact’, Economic
Mobility Corporation, April.
Rodrik, D. (1996), ‘Understanding Economic Policy Reform,’ Journal of Economic Literature, March.
—     (2007), ‘Industrial Policies for the Twenty-First Century’, ch. 4 in D. Rodrik, One Economics, Many
Recipes: Globalization, Institutions, and Economic Growth, Princeton, NJ, Princeton University
Press.
—     (2008), ‘Normalizing Industrial Policy’, Washington, DC, https://drodrik.scholar.harvard.edu/
files/dani-rodrik/files/normalizing-industrial-policy.pdf.
—     (2018), ‘New Technologies, Global Value Chains, and the Developing Economies’, Pathways for
Prosperity Commission Background Paper Series, no. 1, Oxford University.
—     (2019), ‘Towards a More Inclusive Globalization: An Anti-social Dumping Scheme’, Economists
for Inclusive Prosperity Policy Brief 9, February.
—     (2021), ‘Why Does Globalization Fuel Populism: Economics, Culture, and the Rise of Right-Wing
Populism’, Annual Review of Economics, 23, forthcoming.
—     Sabel, C. F. (2021), ‘Building a Good Jobs Economy’, in D. Allen, Y. Benkler, and R. Henderson
(eds), Political Economy and Justice, Chicago, IL, University of Chicago Press, forthcoming.
—     Stantcheva, S. (2021), Economic Inequality and Insecurity: Policies for an Inclusive Economy, Report
for the Macron Commission, June.
Sabel, C. (2007), ‘Bootstrapping Development: Rethinking the Role of Public Intervention in
Promoting Growth’, in V. Lee and R. Swedberg (eds), On Capitalism, Standford, CA, Stanford
University Press, 305–41.
Schaberg, K. (2017), ‘Can Sector Strategies Promote Longer-term Effects? Three-year Impacts from
the WorkAdvance Demonstration’, MDRC Brief, September.
Ton, Z. (2014), The Good Jobs Strategy. Amazon Publishing.